cat sp.txt | text2wave -o sp.wav -eval '(voice_msu_ru_nsh_clunits)'
lame sp.wav sp.mp3
